LangGraph Observability Without Rewriting Your Graph
A single callback exposes every node, LLM call, and tool step.
Tomás Errejón
5 stories
A single callback exposes every node, LLM call, and tool step.
Span monitoring shows the graph ran; only instructions show it worked.
AI agents need monitoring that checks if decisions are correct, not just if systems stay online.
AI agents confidently report success while actually failing silently.
Persona drift in AI agents looks like success until customers quietly leave.